This study analyzes efficiency of weather derivatives as primary insurance instruments for six crop reporting districts that are among the largest producers of corn, cotton, and soybeans in the United States. Specific weather derivatives are constructed for each crop/district combination based on analysis of several econometric models. The performance of the designed weather derivatives is then analyzed both in- and out-of-sample. The primary findings suggest that the optimal structure of weather derivatives varies widely across crops and regions, as does the risk-reducing performance of the optimally designed weather derivatives. Further, optimal weather derivatives required rather complicated combinations of weather variables to achieve reasonable fits between weather and yield.